For large herbivore populations to persist, they must meet needs through each stage of the annual cycle during favourable years, while minimising loss of body stores in adverse years (Parker et al., 2009). Large herbivores must cope with seasonal and inter-annual variation in plant growth and hence in the quantity and quality of food resources, dependent largely on rainfall in savannah ecosystems (Sinclair, 1975; Ellis and Swift, 1988; Owen-smith, 2008), in contrast to temperature at higher latitudes (Mysterud et al., 2001). Temporal variation in resources may be counteracted by exploiting spatial heterogeneity (Scoones, 1995; Owen-Smith, 2002; Owen-Smith, 2004; Fryxell et al., 2005; Stokes et al., 2006; Hobbs et al., 2008; Prins and van Langevelde, 2008), but only if animals can access key components of this heterogeneity. Environmental patchiness or spatial heterogeneity has long been recognised as a key mechanism facilitating access to resources and an optimal diet (MacArthur and Pianka, 1966). Environmental patchiness becomes critical for the survival of a species when different patches provide different non-substitutable resources (Hobbs et al., 2008). For example, moose (Alces alces) in Isle Royale National Park, Michigan, obtain sufficient essential sodium intake (Kaspari, 2020) in one specialised habitat (wetlands), while they maximise their energy intake in woodland habitats (Belovsky, 1978). Adaptive foraging and access to essential non-substitutable resources is facilitated by freedom of movement across broad heterogenous landscapes or ecological gradients supporting diverse functional forage resources (Belovsky, 1978; Senft et al., 1987; Scoones, 1995; Fryxell et al., 2005; Stokes et al., 2006; Hobbs et al., 2008).

Ecological productivity gradients, such as rainfall or hydrological gradients, that influence the spatial distribution of soil nutrients, forage biomass, quality and phenological patterns across the annual cycle determine the distribution of functional seasonal forage resources (McNaughton, 1985; Murray, 1995; Hopcraft et al., 2010; Fynn et al., 2014; Sianga and Fynn, 2017; Keemekae, 2022). In turn, the distribution of these various forage resources influences large herbivore seasonal movements and facilitates adaptive foraging responses to seasonal and inter-annual variability in forage biomass and quality (Bell, 1970; Maddock, 1979; Hopcraft et al., 2010; Fynn et al., 2015). For example, the wetter and higher productivity regions of ecological gradients, such as in high-rainfall areas or in wetlands, support more reliable, adequate-quality (green) forage during the dry season where plant growth ceases and forage becomes scarce and of low quality in drier regions of the gradient (Vesey-FitzGerald, 1960; Sheppe and Osborne, 1971; Acres et al., 1985; McNaughton, 1985; Fryxell and Sinclair, 1988; Pamo, 1998; Kamweneshe, 2000; Scholte and Brouwer, 2008; Fynn et al., 2014; Fynn et al., 2015). A reserve of adequate-quality forage during the dry season is, therefore, essential for reducing loss of body stores of large herbivores and even population collapse during droughts (Illius and O’Connor, 2000; Owen-Smith, 2004). The drier and less productive regions of ecological gradients support shorter and more digestible and nutritious forage (McNaughton, 1985; Fryxell and Sinclair, 1988; Murray, 1995; Hopcraft et al., 2010; Fynn et al., 2014), which is essential for growth and reproduction (White, 1983; Gordon, 1988; Owen-Smith, 1990; Gaillard et al., 2000; Person et al., 2003; Pettorelli et al., 2003; Cook et al., 2004; Owen-Smith, 2004; Parker et al., 2009).

Consequently, for large herbivores to assess and respond to seasonal and inter-annual variability of forage quantity and quality, as well as to variability in their own resource needs, influenced by seasons of growth and reproduction, they need to be able to move across ecological gradients to access functional heterogeneity in forage resources (Owen-Smith, 2004; Hopcraft et al., 2010). Climate change is expected to increase temperatures and environmental variability, especially the frequency and intensity of droughts (Cai et al., 2014; Engelbrecht et al., 2015). Higher temperatures will increase disparity between rainfall and evapotranspiration leading to shorter growing seasons and thereby a shorter duration of green forage for large herbivores; this will likely make it more difficult for large herbivores to meet their nutritional requirements over the annual cycle (Ogutu and Owen-Smith, 2003). The range of different forage resources with distinct functional roles that enable adaptation to environmental variability and climate change is more nuanced and diverse than the simple dichotomy of wet and dry season forage resources.

Thus, knowledge of the full range and function of various forage resource classes and how they are distributed on ecological gradients is essential for understanding how large herbivores assess and respond to seasonal and inter-annual variability in forage quantity and quality, as well as to climate change. Such an understanding will play a key role in planning for the effects of climate change and in designing protected areas (PAs). Anatomical traits of herbivores, such as body size and mouth anatomy, can strongly influence how various large herbivore species perceive the functionality of different types of forage resources. Body size is considered a key trait influencing niche differentiation along forage quality gradients (Owen-Smith, 1988; Hopcraft et al., 2010). While body size does not appear to influence mean retention time of forage and digestive efficiency (Müller et al., 2013; Steuer et al., 2014), there are several mechanisms through which larger bodied herbivores can be more tolerant of lower quality forage. First, larger body size may promote greater ability to handle and process low-quality forage (Steuer et al., 2014). Second, larger body size may allow greater forage intake relative to basal metabolism, allowing large-bodied herbivores to survive better on lower quality forage (Müller et al., 2013). Another key factor that may promote coexistence of large herbivores on forage quality gradients is that with low quality forage, where lignin content is high, a longer retention time of forage in the gut will not be as advantageous for nutrient extraction than having a higher intake rate of forage (Müller et al., 2013). Therefore, hindgut fermenters are likely to be more tolerant of low-quality forage than ruminants because their intake and passage rates of low-quality forage is greater than that of ruminants (Müller et al., 2013). Body size also influences predation risk with population size of very large-bodied herbivores (buffalo and upwards) being unconstrained by predation (Sinclair et al., 2003; Hopcraft et al., 2012).

Perhaps of greater significance than body size for coexistence on forage quality gradients is mouth anatomy, which influences intake rates of forage in different grassland structural states where broad-mouthed grazers can achieve sufficient bite size to meet intake requirements in short grassland with the reverse being true for narrow-mouthed grazers in taller grassland (Murray and Illius, 2000; Arsenault and Owen-Smith, 2008). Finally, adaptive resilience in large herbivore populations will be determined by their ability to assess and respond to predation risk, which if compromised by anthropogenic effects, can cause the collapse of herbivore populations (Bergerud and Elliot, 1986; Harrington et al., 1999; Rettie and Messier, 2000; Wittmer et al., 2005). Understanding how forage resource classes and anatomical traits are related to predation risk is essential for understanding the adaptive foraging strategies of large herbivores and how these may by disrupted by anthropogenic effects. The anatomical traits of large herbivores, therefore, provides a mechanism for coexistence by facilitating partitioning of forage resources, as well as affecting predation avoidance strategies.

Another important concept not yet integrated into our understanding of functional heterogeneity of forage resources and adaptive foraging strategies is the concept of medicinal resources. Most discussion has centred around energy and protein in forage and predation risk as drivers of adaptive foraging movements (e.g. Parker et al., 2009; Hopcraft et al., 2010). Wild herbivores face the same challenges from internal parasites, ticks and disease as domestic ones, whose populations in confinement are maintained through veterinary inputs in the form of synthetic drugs and antibiotics (Acha and Melendez, 1983). However, when domestic herbivores have access to high plant and phytochemical diversity (secondary metabolites) across broad landscapes, they have much less need for veterinary inputs (see Provenza et al., 2003; Villalba et al., 2006; Provenza et al., 2007; Villalba and Provenza, 2007; Provenza and Villalba, 2010; Gessner et al., 2017; Villalba et al., 2017; Provenza, 2018). It follows, therefore, that with wild herbivores having no access to veterinary interventions, their populations, together with genetic adaptation to disease through natural selection, must rely on plant secondary metabolites to remain healthy and productive. In this light, we suggest that various medicinal resources in the form of a diverse array of plant secondary metabolites (phytochemical diversity) represents an essential aspect of functional resource heterogeneity for large herbivores but is yet to be appropriately integrated into a foraging ecology framework.

In summary, for large herbivores to maintain viable populations and to coexist with other sympatric large herbivore species requires assessment of and response to various abiotic and biotic factors, with responses being mediated by various contingency factors ( Figure 1 ). In this paper, we synthesise and describe the nature and interactions of the various components of Figure 1 . While functional forage resource classes have been developed for African browsers (Owen-Smith, 2002), a similar framework for African grazers is lacking, especially in understanding how these resources are distributed on ecological productivity gradients. Moreover, while much attention has been given to body size as a key anatomical trait of large herbivores affecting foraging and anti-predation strategies, much less attention has been given to mouth anatomical traits. Understanding is needed on the link between mouth anatomy and foraging strategies and how these determine anti-predation strategies. Thus, there is a gap in our understanding of the biotic component of Figure 1 and how it relates to the abiotic and contingency components. We, therefore, focus on developing an understanding of the range of functional forage resource classes that large herbivore grazers in African savannas require to respond to environmental variability and disease and how these resources are distributed on ecological gradients. In addition, we show how large herbivore anatomical traits influence their perceptions of the functionality of these forage resources and how this influences predation avoidance strategies and overall species coexistence.

Decades of detailed observations on diet selection by kudu Tragelaphus strepsiceros and several other browser species enabled linking of the physical and chemical traits of plants to key adaptive foraging strategies over the annual cycle (Owen-Smith et al., 1983; Owen-Smith and Cooper, 1987; Cooper et al., 1988; Owen-Smith and Cooper, 1989; Owen-Smith, 1994). This enabled empirical analysis and modelling of the influence of different forage resource classes on population dynamics (Owen-Smith, 1990; Owen-Smith, 2002, p 255). For browsers, key plant traits determining seasonal adaptive foraging were protein content - condensed tannin ratios (nutrient supply vs. chemical defences), physical defences (thorns) and seasonal availability of forage determined by phenological traits (see Owen-Smith et al., 1983; Owen-Smith and Cooper, 1987; Cooper et al., 1988; Owen-Smith and Cooper, 1989; Owen-Smith, 1994). Linking these plant traits to adaptive foraging strategies was achieved by developing six generic functional resource classes (Owen-Smith, 2002, p 255): high-quality resources as soft leaves and stems of digestible, protein-rich forbs; restricted-intake resources, such as protein-rich leaves of thorny species, where thorns limit intake; staple resources as leaves of palatable deciduous species that provide a more reliable intake of relatively protein-rich forage, low in tannins; reserve resources as leaves of palatable evergreens that provide adequate-quality forage during the dry season when more preferred forages are depleted; transient bridging resources as palatable emerging (unpalatable when mature), relatively protein-rich leaves on unpalatable deciduous trees during the late dry season; and low-quality buffer resources as leaves of unpalatable evergreens. Generic functional resource’s enable browsers to meet requirements for growth and reproduction, while minimising body store depletion during the dry season.

Building on the generic functional resource concept we develop a forage resource – ecological gradient framework for grazers in African savannas, which we refer to as functional adaptive resources (FARs). We propose seven FAR classes in African grazing ecosystems that are distributed across soil moisture ( Figure 2 ) and disturbance ( Figure 3 ) gradients.

When considering FARs for large herbivores it is essential to recognise that their body size, digestive anatomy and mouth anatomy greatly affects the functionality of different resource types for maximising intake rates (Murray and Illius, 2000; Arsenault and Owen-Smith, 2008; Codron et al., 2008; Hopcraft et al., 2012). Thus, selection of specific resource classes and habitat types is dependent on abiotic and biotic factors determining the types of resources and predation risk in combination with the contextual factors of anatomical traits and environmental variation, as well as anthropogenic impacts ( Figure 1 ). Body size is an important factor determining selection for different grassland structural states because greater intake requirements with increasing body size necessitates consumption of more abundant lower quality forage to meet their intake requirements (Hopcraft et al., 2010; Sensenig et al., 2010; Müller et al., 2013; Steuer et al., 2014). However, specialised mouth adaptations may greatly modify intake rates of large herbivores on different grass heights (Murray and Illius, 2000), resulting in mouth anatomy being a key factor determining selection for grass height (Arsenault and Owen-Smith, 2008). Narrow-mouthed large herbivores (beyond a certain body size) cannot maintain sufficient intake rates on short grass, with the reverse being true for broad-mouthed large herbivores on tall grass (Murray and Illius, 2000).

Mouth anatomy (together with body size) determines the height of grass considered as a fecundity resource during the wet season for the large herbivore species in question. Narrow-mouthed grazers, such as roan and sable antelope, waterbuck (Kobusellipsiprymnus) and reedbuck (Redunca arundinum), can be more selective of plant parts, selecting fresh digestible leaves on medium height and tall grasses while avoiding unintentional intake of indigestible stems and low-quality dead leaves (Jungius, 1971; Grobler, 1981; Schuette et al., 1998; Murray and Illius, 2000; Knoop and Owen-Smith, 2006; Kassa et al., 2007; Havemann et al., 2022). During the wet season, roan and sable antelope eat high-quality leafy grasses of medium height, such as Andropogon spp., Panicum maximum, Panicum coloratum, Digitaria eriantha and Andropogon and Hyparhenia spp (Grobler, 1981; Schuette et al., 1998; Knoop and Owen-Smith, 2006; Hensman et al., 2013; Havemann et al., 2022). Thus, these medium height leafy grasses, as opposed to lawn grasses of short grass grazers, can be fecundity resources for narrow-mouthed grazers. The greater selective capacity of narrow mouthed grazers enables them to switch to very tall grasses (>1.8m) growing in drainage sumps and to browse in woodlands during the dry season facilitating intake of green forage when most grasses have dried out (Schuette et al., 1998; Knoop and Owen-Smith, 2006; Kassa et al., 2007; Hensman et al., 2012; Selebatso et al., 2018). Trees and shrubs have much deeper root systems than grasses, enabling them to access soil moisture in deeper layers and maintain green leaves much later into the dry season than grasses. Very tall grasses growing in poorly drained sumps (wet meadows, vleis, dambos) also retain green leaves much later into the dry season than short grasses (Knoop and Owen-Smith, 2006; Fynn et al., 2014). Most narrow-mouthed tall grass grazers it appears are non-migratory, often living in woodlands all year round (Jungius, 1971; Grobler, 1981; Schuette et al., 1998; Murray and Illius, 2000; Knoop and Owen-Smith, 2006; Kassa et al., 2007), where the only source of green forage in the dry season is the leaves of woody species or on very tall grasses in poorly drained sumps.

Thus, having a narrow mouth is a key adaptation for being able to forage effectively in lower quality taller grass habitats in backcountry woodlands, thereby avoiding competition with high densities of short grass grazers in high-quality short grasslands. Short grasslands are generally favoured by wildebeest, gazelles, impala, warthog and zebra (Arsenault and Owen-Smith, 2008; Bhola et al., 2012). Thus, variation in mouth anatomy provides a key mechanism facilitating coexistence among a guild of large herbivore species by enabling spatial partitioning of foraging sites across landscapes. In addition, by restricting themselves to lower quality tall grasslands in backcountry woodlands, narrow-mouthed grazers avoid high densities of predators associated with high densities of grazers (Harrington et al., 1999).

By contrast, the broad-mouths of short grass grazers, such as wildebeest, enable greater bite size on short grasses than narrow-mouthed grazers (Murray and Illius, 2000). However, their inability to forage effectively on tall grasses and browse (Murray and Illius, 2000), constrains broad-mouthed grazers to foraging dominantly on shorter grasses throughout the year with much less browse intake during the dry season than narrow-mouthed grazers (Selebatso et al., 2018). Consequently, broad-mouthed grazers often migrate to floodplains or high-rainfall areas during the dry season where they can find grasses retaining green leaf (Maddock, 1979; Fynn et al., 2014; Naidoo et al., 2014; Bennitt et al., 2016; Sianga et al., 2017a). With broad-mouthed grazers requiring short, high-quality forage they favour fresh regrowth of grasses after fire or grazing (Vesey-FitzGerald, 1960; Sensenig et al., 2010; Hempson et al., 2015). Spatial partitioning of foraging sites due to mouth anatomy is also observed with woodland caribou (Rangifer tarandus) in Boreal regions. Woodland caribou have specialised mouth adaptations to forage on lichens in old growth forests, thereby avoiding much overlap of moose, and their associated predators (wolves), where moose favour browse on regrowth in clear felled forests (Bergerud and Elliot, 1986; Rettie and Messier, 2000; Wittmer et al., 2005). It appears, therefore, that mouth anatomy plays a critical role in niche diversification and coexistence among large herbivores in both African and North American ungulates.

Several anatomical innovations result in divergences from the typical seasonal foraging patterns of narrow and broad mouthed grazers. While buffalo have a broad mouth, their very large body size combined with use of their tongues to increase bite size, necessitates that instead of short grasses being favoured, they prefer medium height leafy grasses in areas with lower concentrations of grazers (Bhola et al., 2012). Consequently, buffalo favour the same backcountry woodland habitats as roan and sable antelope during the wet season (Vesey-FitzGerald, 1960; Taylor, 1985; Prins, 1996; Bennitt et al., 2016; Sianga et al., 2017a). With their broad mouths, however, buffalo cannot forage effectively for green leaf on tall grasses and browse in woodlands during the dry season so they retreat to floodplains at this time (Vesey-FitzGerald, 1960; Bennitt et al., 2016; Sianga et al., 2017a). In addition, body size can determine predation risk for a large herbivore, which can influence habitat selection for large herbivores smaller than buffalo (Sinclair et al., 2003; Hopcraft et al., 2012; Yoganand and Owen-Smith, 2014). The implication is that a habitat type may be essential for the viability of a large herbivore population, not only through its specific nutritional resources but also by minimising predation risk. For example, wildebeest benefit from open, short grass habitats, not only because they provide higher forage quality than tall grass habitats but also because they provide much better visibility for the detection of predators (e.g. Yoganand and Owen-Smith, 2014). The small-bodied antelope, Oribi (Ourebia ourebi), forage in short grass habitats for the better forage quality and visibility that they provide but rest in tall grass habitats to avoid detection (Everett et al., 1991). Wildebeest drop their calves in open, high visibility habitats, whereas sable and roan antelope hide their calves in tall grass and scrub far from areas with high densities of predators. Large body size also facilitates greater ability to survive on low-quality forage by being more able to cope with processing coarse forage (having a larger mouth and chewing ability) and by having a lower metabolic rate relative to food intake (Müller et al., 2013).

Thus, body size and mouth anatomy are critical morphological adaptations that enable a wide range of different foraging strategies and predator avoidance mechanisms across landscapes and seasons. There appear to be three main foraging and predator avoidance strategies of large herbivore species over the annual cycle. First, those with specialised broad mouths (e.g. wildebeest) or small enough body size (e.g. impala and gazelles) favour habitats with high visibility, thereby minimising predation risk: they rely on early detection and flight from predators. Second, those with narrow mouths (but greater body size than gazelles) can forage in tall grass habitats and woodlands far from high concentrations of short grass grazers where predator densities are high: they rely on spatially avoiding predators. Their ability to use browse effectively in the dry season enables them to remain all year round in these habitats (Hensman et al., 2013; Havemann et al., 2022). Third, those with a tongue sweep strategy for increasing bite size that constrains them to foraging on medium height leafy grasses in woodlands far from high concentrations of short grass grazers (e.g. buffalo). Their large body size and defensive abilities in large herds enables them to forage in woodlands and scrub where visibility is low: they rely on fending off predators. Another anatomical trait that also promotes niche differentiation on forage quality gradients is gut anatomy, with hind gut fermenters being able to ingest more low-quality forage than ruminants, which outweighs any advantage of longer retention times on digestive efficiency (Müller et al., 2013). These different anatomical traits account, to a large degree, for coexistence of the wide diversity of large herbivore species across heterogeneous African savanna landscapes. However, considering the emphasis that has been placed on body size in the Jarman-Bell hypothesis, it appears that mouth anatomy has been greatly neglected as a mechanism of coexistence and in can even override the effect of body size as a determinant of grass height selection e.g. with white rhino (Arsenault and Owen-Smith, 2008).

In this synthesis we attempt to provide an understanding of the range of functional forage resource classes and medicinal resources (FARs) required by large herbivore grazers over the annual cycle in the context of their anatomical traits ( Figure 1 ). Additionally, we develop a framework linking the FAR classes to ecological productivity gradients at various scales determined by spatial variation in soil moisture availability on topographic, rainfall or hydrological gradients at various scales ( Figure 3 ), as well as by spatial variation in grazing intensity ( Figure 3 ). We show how large herbivores rely on a diverse range of functional resource classes (FARs) to enable them to respond to seasonal and interannual variation in forage quantity and quality and to their own variation in intake requirements for various resources. Consequently, reducing access by large herbivores to any class of FARs will negatively impact their populations (Illius and O’Connor, 2000; Owen-Smith, 2002; Owen-Smith, 2004). We also introduce the concept of medicinal resources as a critical functional adaptive resource class that has been all but ignored in large herbivore foraging ecology theory. We show that productive and healthy herbivore populations are dependent upon access to a wide diversity of plants and the associated phytochemicals that they provide for therapeutic and prophylactic defence against toxins, parasites and disease.

We also show how the perception of the functionality of these resources by various large herbivore species is determined by their anatomical traits, such as body size, mouth anatomy and digestive anatomy, which also determines predation avoidance strategies. Body size determines vulnerability to predation, whereas various combinations of body size, digestive anatomy and mouth anatomy determine selection for grass height and also predation avoidance strategies. Although body size is an important trait to account for, mouth anatomy can play a dominant role in determining selection for different grassland structural states. For example, while larger body size generally results in selection for a taller grass, lower quality diet, the exceptionally broad mouth of the very large bodied white rhino enables it to forage effectively on short grazing lawns (Arsenault and Owen-Smith, 2008). By contrast, the narrow mouth of the small-bodied reedbuck enables it to forage effectively in tall grassland (Jungius, 1971). Consequently, niche diversification and large herbivore species coexistence is determined by the various combinations of body size, digestive anatomy and mouth anatomy.

Access to the full range of FARs along environmental productivity gradients is being compromised by anthropogenic effects, such as ecosystem fragmentation and climate change. Lack of access to reserves of forage in the more productive regions of ecological gradients can result in declining body stores and starvation during the dry season and droughts, leading to population collapse (Walker et al., 1987; Ellis and Swift, 1988; Illius and O’Connor, 2000; Owen-Smith, 2004). Thus, the wettest end of productivity gradients is critical for preventing starvation during droughts (a buffer or key resource), thereby pointing to the importance of these wet zones for adaptation to predicted increased temperature and drought frequency under climate change (Cai et al., 2014; Engelbrecht et al., 2015). Lack of access to high-quality (fecundity) resources in the less productive regions of ecological gradients compromises growth and reproduction, which can result in declining population size (Gaillard et al., 2000; Owen-Smith, 2004; Parker et al., 2009). It becomes clear, therefore, that wet season fecundity resources should also be considered as key resources, not just dry season reserve and buffer resources, because large herbivore populations cannot attain a large population size without access to fecundity resources. In addition, without access to appropriate plant diversity distributed across ecological gradients, the health and productivity of large herbivore populations is expected to suffer. The most productive and stable herbivore populations will, therefore, be those with access to the full range of FARs on ecological productivity and disturbance gradients.

Fecundity in large herbivore populations is determined not only by sufficient intake of energy and protein, but also by sufficient intake of key minerals. With minerals such as Ca, Na and P being essential for pregnancy and lactation, as well as for several essential physiological processes (Kaspari, 2020), most known migrations in Africa are to saline regions for the wet season (Maddock, 1979; Breman and De Wit, 1983; Joos-VandeWalle, 2000; Bartlam-Brooks et al., 2011; Fynn and Bonyongo, 2011; Bradley, 2012; Fynn et al., 2014; Naidoo et al., 2014). Non-migratory large herbivores use patchy saline areas in the landscape during the wet season, such as sodic sites and nutrient hotspots (McNaughton, 1988; Grant and Scholes, 2006; Anderson et al., 2010; Augustine et al., 2011). Saline zones of ecological gradients support the lowest plant productivity along these gradients, having the shortest growing season and the lowest biomass production ( Figure 4 ). As conditions progressively dry out after the last rains and fecundity resources and drinking water become depleted, large herbivores tend to move up the productivity gradient as the dry season progresses, selecting taller lower quality grasses and sedges, which function as ungrazed reserves of forage (Bell, 1970; Fryxell and Sinclair, 1988; Hopcraft et al., 2010; Fynn et al., 2015). Large herbivores increasingly forage in deeper-flooded zones of wetlands as the dry season proceeds (Vesey-FitzGerald, 1960; Sheppe and Osborne, 1971; Tinley, 1977; Kamweneshe, 2000), especially during droughts (Western, 1973). Similarly, wildebeest migrate further up the rainfall gradient in Serengeti in dry years than in wetter years (Maddock, 1979).

With these observations as a background, our FAR-ecological gradient framework has important implications for research and for conservation and development planning across broad landscapes, especially in the context of climate change (Cai et al., 2014; Engelbrecht et al., 2015) and accelerating fragmentation of landscapes (Ogutu et al., 2011; Morrison and Bolger, 2014; Løvschal et al., 2017; Veldhuis et al., 2019). For example, data on topography, soils, hydrology, rainfall and remote-sensed seasonal NDVI can be modelled to produce soil moisture/productivity gradients in a GIS, allowing the development of a spatially explicit FAR resource map. FAR mapping can be used to guide development of new PAs or upgrades of existing PAs, by assessing the optimal spatial configuration that a PA requires to include the full range of FARs needed to maintain productive and stable large herbivore populations. FAR mapping will also provide an objective approach to habitat suitability assessments for different large herbivore species, allowing an assessment on whether certain large herbivore species can be introduced to a conservation area and where certain forage resource deficits may occur.

Pereira and Gregorini (2022) show how remote sensing and GIS can be used to assess the spatial distribution of phytochemically-rich plants thus enabling people to visualize potential healthscapes for wild or domestic herbivores. The maps they created using advanced image classification of botanical composition and phytochemicals show the distribution of phytochemicals in plants in ways that enable managers to make better grazing management decisions for more productive, sustainable, and healthy grazing systems. Assessment of healthscapes is a paradigm shift in rangeland management thinking with potential applications for conservation planning, such as in determining the optimal spatial configuration of PAs.

In addition, modelling can be used to assess the role that different FAR classes play in determining the productivity and stability of large herbivore populations, such as what would happen if access to certain FAR classes was lost through ecosystem fragmentation, or if drought frequency was increased under climate change or how the wet end of ecological gradients mediates the impacts of drought (e.g. Illius and O’Connor, 2000; Owen-Smith, 2002, p255; Owen-Smith, 2004). Modelling could also assess how management options, such as implementing fire management or strategic livestock grazing to stimulate regrowth of short, high-quality forage in drylands during the wet season (a non-saline fecundity resource), or in wetlands or high-rainfall regions during the dry season (a bridging resource), would affect large herbivore population productivity, after accounting for large herbivore anatomical traits (e.g. Gordon, 1988; Fynn et al., 2016). Increasing the coverage of short grassland may benefit broad mouthed grazers, such as wildebeest and white rhino, but will likely negatively impact narrow-mouthed grazers (Harrington et al., 1999; Murray and Illius, 2000; Arsenault and Owen-Smith, 2008; Bhola et al., 2012). It is, therefore, essential to consider herbivore anatomical traits to understand how they perceive and respond to various FAR classes and to predation risk (Harrington et al., 1999; Murray and Illius, 2000; Arsenault and Owen-Smith, 2008). For example, populations of narrow-mouthed, tall grass grazers, such as roan and sable antelope in Kruger National Park collapsed after placement of artificial water points in backcountry waterless regions far from perennial rivers (Harrington et al., 1999). This unintended consequence of management could have been avoided by consideration of how FARs are distributed on grazing gradients determined by distance to perennial water (Sianga et al., 2017b; Figure 3 ). Artificial water points may disrupt these landscape-scale grazing gradients, with various consequences for large herbivores with different anatomical traits (Harrington et al., 1999; Sianga et al., 2017b).

We provide a forage resource-environmental gradient framework for understanding how large herbivores in African savannas assess and respond to variability in forage resources and their seasonal demands for resources, as well as to predators, parasites and disease. We demonstrate that various large herbivore foraging responses and predation avoidance strategies are mediated by their anatomical traits, which also facilitates their coexistence. We highlight that mouth anatomy has been neglected as a key anatomical trait determining large herbivore selection for grass height. Our framework provides a basis for refining and developing our understanding of factors affecting conservation of diverse guilds of large herbivores in African savannas under landscape fragmentation and climate change.

RF conceptualised and developed the nutrition aspects of functional adaptive resources and the mechanisms through which they are distributed on ecological gradients. FP conceptualised and developed the medicinal aspects of functional adaptive resources. Both authors contributed towards the writing and editing of the manuscript. All authors contributed to the article and approved the submitted version.